Power transformer health index calculation method based on principal component analysis and mass-spring-damping model

Abstract

The invention discloses a power transformer health index calculation method based on principal component analysis and a mass-spring-damping model. The method comprises the following steps: collecting multi-stage data of analysis parameters and oil quality parameters of dissolved gas in transformer oil; determining a weight coefficient of each parameter by using a principal component analysis method; calculating a DGA index and an oil quality index in combination with a mode coefficient in a CIGRE technical manual 761; enabling the transformer to be equivalent to a second-order mass-spring-damping system, and mapping a DGA index and an oil quality index into a damping coefficient and an inherent frequency parameter of the system respectively; and constructing a system state space matrix, and calculating a system matrix infinite norm ratio of the new transformer to the transformer in operation to obtain a transformer health index. The method improves the accuracy and reliability of health index calculation, and is suitable for transformer state evaluation and life prediction.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power equipment state monitoring and fault diagnosis, and particularly relates to a calculation method of a power transformer health index. BACKGROUND

[0002] The power transformer is one of the most important devices in the power system, and its operation state directly affects the reliability and safety of the power grid. The transformer is subjected to the combined action of electrical stress, thermal stress and chemical reaction in operation, resulting in insulation aging and performance degradation. At present, the commonly used transformer state evaluation methods include dissolved gas analysis, oil quality parameter detection and paper insulation polymerization degree measurement. In the prior art, the health index is calculated based on single measurement data or fixed weight coefficient, ignoring the parameter change trend and the coupling relationship between multiple parameters, resulting in insufficient accuracy of the evaluation result. In addition, the traditional method does not combine the dynamic characteristics of the transformer with the mechanical system model, and it is difficult to comprehensively reflect the health state of the transformer. SUMMARY

[0003] In the transformer state evaluation, how to: ① objectively use multi-period monitoring data to reduce accidental errors; ② incorporate the physical mechanism of oil-paper insulation degradation into the index; ③ provide a health index with consistent scale, 0-1 range and easy horizontal comparison, is a problem to be solved.

[0004] In order to achieve the above technical purpose, the technical scheme of the present application is as follows:

[0005] A power transformer health index calculation method based on principal component analysis and mass-spring-damper model, comprising:

[0006] S1) Multi-period data acquisition and grading: obtaining dissolved gas analysis data and oil product electric-matter-chemical parameters of a target transformer in several sampling periods; grading each parameter according to the color-coefficient mode of CIGRE Technical Manual 761 and assigning a mode coefficient;

[0007] S2) Weight determination: applying principal component analysis to the data of the several sampling periods to obtain the weight of each variable, which is used to reflect the change trend of the parameter in the time sequence;

[0008] S3) Index construction: weighting the dissolved gas analysis data and oil quality according to the weight to obtain the dissolved gas analysis data index and oil quality index respectively, which are valued in the interval [0, 1];

[0009] S4) Physical mapping and modeling: mapping the dissolved gas analysis data index and oil quality index into the damping coefficient and natural frequency parameters of the second-order mass-spring-damper model respectively, and establishing a second-order system state space matrix representing the state of the transformer;

[0010] S5) Health Index Output: Calculate the infinity norm of the state space matrix of the "new" transformer and the in-service transformer respectively, and take the ratio of the two as the health index HI.

[0011] The weight of the dissolved gas analysis data index and the oil quality index is obtained from multiple sampling period data based on principal component analysis to reduce the evaluation error caused by using only a single measurement.

[0012] The weight calculation of principal component analysis uses the inverse of the variance of each parameter as the weight, and selects the minimum principal component of each parameter to participate in the calculation.

[0013] The dissolved gas analysis data index is obtained by weighted average of the mode coefficient of each gas and the corresponding principal component analysis weight, and the oil quality index is obtained by weighted average of the mode coefficient of the most unfavorable value of each parameter of the oil and the corresponding principal component analysis weight.

[0014] The mode coefficient of the most unfavorable value of the oil is the CIGRE 761 mode color-coefficient corresponding to the worst sample of each sampling period.

[0015] The dissolved gas analysis data index is mapped to the damping coefficient of a second-order system, and the oil quality index is mapped to the natural frequency parameter, and a state space matrix is constructed accordingly.

[0016] The health index HI is obtained by calculating the ratio of the infinity norm of the new transformer system matrix to the infinity norm of the in-service transformer system matrix.

[0017] The conversion relationship of furfural-paper insulation polymerization degree in CIGRE 761 is used to determine or correct the state space model parameters to reflect the influence of paper insulation degradation on the health index:

[0018] ,

[0019] The dissolved gas analysis data component at least includes ; The oil quality parameters at least include water content, acid value, furfural, breakdown voltage, dielectric loss factor and interfacial tension.

[0020] The value range of the index is [0, 1]; The smaller the value, the lower the degree of degradation of the corresponding factor on the performance, and the closer the health state to the standard "new" state.

[0021] The present application significantly improves the accuracy and reliability of the health index calculation through multi-period data trend analysis and mechanical system simulation, and is suitable for transformer state evaluation, life prediction and maintenance decision. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 is a flowchart of the method of the present application;

[0023] Figure 2 A schematic diagram for principal component analysis method. DETAILED DESCRIPTION

[0024] The application will be further described in conjunction with the accompanying drawings of the specification:

[0025] The application relates to a power transformer health index calculation method based on principal component analysis and a mass-spring-damper model, and specifically comprises the following steps:

[0026] Step 1: On the same in-service transformer, covering at least three sampling periods, collect dissolved gas analysis (DGA: ) and oil electro-physical and chemical parameters (including water, acid value, furfural, breakdown voltage BDV, dielectric loss factor tan delta, interfacial tension IFT, etc.); perform time alignment, missing value completion and abnormal value elimination (such as judging by median number plus or minus 3 times MAD).

[0027] Step 2: According to the CIGRE 761 standard, map each parameter to the A-E level according to the interval, and assign the mode coefficient {0, 1, 2, 4, 10} to obtain the mode coefficient matrix of “multi-period x multi-variable”.

[0028] Step 3: PCA weight determination: execute principal component analysis (PCA) with the matrix obtained in step S2 as input to obtain the variable weight vector (W); the combination strategy of “variable variance reciprocal x minimum principal component load” can be used to enhance the robustness to occasional abnormalities, and the weight is normalized.

[0029] Step 4: Calculate two types of dimensionless indexes and linearly normalize them to [0, 1]:

[0030] ,

[0031] Step 5: Estimate the degree of polymerization of paper insulation from the furfural concentration

[0032]

[0033] Define the equivalent mass m = DP / 1200 (new paper insulation m = 1 approximately).

[0034] Step 6: Map DGA to damping ratio ( ), map OQ to inherent frequency related quantity ( ), and accordingly obtain damping / stiffness parameters

[0035] ,

[0036] Step 7: Construct the second-order state space matrix of the in-service transformer

[0037]

[0038] Define "new machine" matrix with DGA=0, OQ=0, m=1

[0039]

[0040] Step 8: Calculate with infinity norm With , output health index

[0041]

[0042] Where HI closer to 1 means healthier.

[0043] Step 9: Recalculate steps 3-8 at fixed period (e.g. week / decade / month) in monitoring master station or edge gateway; set alarm thresholds (e.g. HI<0.6 as attention, HI<0.4 as early warning), and combine trend (ΔHI / Δt) to judge degradation rate.

[0044] Step 10: System consists of data acquisition and classification module, PCA weight calculation module, index construction module, equivalent mass calculation module, parameter mapping module, state space and norm calculation module, health index output module; modules pass through standardized data interface (CSV / JSON / OPC-UA) transmission, output HI and operation and maintenance platform linkage, realize early warning push, repair priority sorting and life assessment report generation.

[0045] The following will be described in detail through a specific example:

[0046] Take a transformer as an example, collect multi-period DGA and oil quality data, including Gas content, and breakdown voltage, dielectric loss factor, acid value, interfacial tension, moisture and furfural content, etc.

[0047] Calculate the weight of each parameter using principal component analysis, and select the minimum principal component as the weight coefficient. Combine the CIGRE 761 mode coefficient to calculate the DGA index and oil quality index. The transformer is equivalent to a mass-spring-damping system, where:

[0048] · Mass represents the degree of polymerization of paper insulation;

[0049] · Spring constant ;

[0050] · Damping coefficient .

[0051] Table 1 shows the gases measured in the transformer, which are the maximum amount of each gas at the time of sampling according to the CIGRE Brochure 761. In calculating the DGA index, the modal coefficient (the highest value) of the colored gases in Table 1 and the weight coefficient obtained by the PCA method are used.

[0052]

[0053] where G is the modal coefficient of the maximum value of each gas at the time of sampling.

[0054] Table 1 Color coding of DGA data of each sampling period of the transformer

[0055]

[0056] The oil quality index is obtained by the following equation

[0057]

[0058] where Q is the modal coefficient of the worst value of the electrical, physical, and chemical parameters of the transformer oil. For example, the modal coefficient of the sample including the lowest breakdown voltage of the oil or the modal coefficient of the largest amount of acidity, moisture, or furfural component within the sampling period. Table 3 shows the electrical, physical, and chemical parameters measured in the transformer oil.

[0059] Table 2 Color coding of transformer oil data of each sampling period

[0060]

[0061] As shown in Figure 2 , the two-dimensional data is reduced to one dimension using principal component analysis. The data of Tables 2 and 3 is reduced from the previous 11 dimensions to 4 dimensions using principal component analysis, as shown in Table 3. In Table 3, the positive and negative signs of the principal components indicate the distribution direction of each component in different quadrants of the coordinate plane.

[0062] Table 3 Weighting factors for reducing data from 11 dimensions to 4 dimensions using the PCA method of the transformer

[0063]

[0064] In Table 4, the transformer health index evaluation method proposed in this paper is compared with some other methods. In other methods, data of one measurement stage is used according to the weight factor and score related to the value of each data. But in this method, the measurement data of multiple stages (trend of data) is used, and the weight factor is obtained from the CIGRE Brochure 761 mode based on the PCA method, rather than the traditional score. As can be seen from Table 4, the proposed method is more accurate than the compared methods.

[0065] Table 4 Method comparison​

[0066]

Claims

1. A method for calculating a health index of a power transformer based on principal component analysis and mass-spring-damper model, characterized in that, Comprise: S1) Multi-period data collection and grading: Obtain the dissolved gas analysis data and oil quality parameters of the target transformer in multiple sampling periods; Grading and assigning mode coefficients to each parameter according to the mode of CIGRE Technical Manual 761; S2) Weight determination: Apply principal component analysis to the data in the multiple sampling periods to obtain the weight of each variable, which reflects the change trend of the parameter in the time series; S3) Index construction: Weight the dissolved gas analysis data and oil quality according to the weight to obtain the dissolved gas analysis data index and oil quality index respectively, which are valued in the interval [0,1], and the calculation formula is as follows ， ， where G i is the pattern coefficient of each gas; Q j is the pattern coefficient of the "worst value" of each oil parameter in each sampling period; W i and W j are the i / jth weight coefficients obtained by the PCA method, respectively; S4) Physical mapping and modeling: Map the DGA index and OQ index to the damping coefficient and natural frequency parameters of the second-order mass-spring-damper model respectively, and establish the second-order system state space matrix representing the state of the transformer; S5) Health index output: Calculate the infinity norm of the state space matrix of the "new" transformer and the in-service transformer respectively, and take the ratio of the two as the health index HI.

2. The method of claim 1, wherein, The weights of the dissolved gas analysis data index and the oil quality index are obtained based on principal component analysis from at least 5 sampling period data to reduce the evaluation error caused by using single measurement.

3. The method according to claim 1 or 2, characterized in that, The weight calculation of principal component analysis uses the inverse of the variance of each parameter as the weight, and selects the minimum principal component of each parameter for calculation.

4. The method of claim 1, wherein, The dissolved gas analysis data index is obtained by weighted average of each gas mode coefficient and the corresponding principal component analysis weight, and the oil quality index is obtained by weighted average of the mode coefficient of the most unfavorable value of each oil parameter and the corresponding principal component analysis weight.

5. The method of claim 4, wherein, The mode coefficient of the most unfavorable value of the oil is the CIGRE 761 mode color-coefficient corresponding to the worst sample in each sampling period.

6. The method of claim 1, wherein, The dissolved gas analysis data index is mapped to the damping coefficient of the second-order system, and the oil quality index is mapped to the natural frequency parameter, and the state space matrix is constructed accordingly.

7. The method of claim 1, wherein, The health index HI is obtained by calculating the ratio of the infinity norm of the new transformer system matrix to the infinity norm of the in-service transformer system matrix.

8. The method of claim 1, wherein, The conversion relationship of furfural-paper insulation polymerization degree in CIGRE 761 is used to determine or correct the parameters of the state space model to reflect the influence of paper insulation degradation on the health index: 。 9. The method of claim 1, wherein, The dissolved gas analysis data set comprises at least ; the oil quality parameters at least comprise water content, acid value, furfural, breakdown voltage, dielectric loss factor and interfacial tension.

10. The method of claim 1, wherein, The value range of the index is [0,1]; the smaller the value, the lower the degradation of the corresponding factor on the performance, and the closer the health state to the standard "new" state.

Images